An inflammation-based cumulative prognostic score system in patients with diffuse large B cell lymphoma in rituximab era

Diffuse large B-cell lymphoma (DLBCL) is the most common subtype of non-Hodgkin lymphoma (NHL), representing 30–40% of all lymphomas. It is an aggressive lymphoma with heterogeneous clinicopathological, immunephenotypic, genetic features and various clinical outcomes [1–4]. Although the addition of rituximab (R) to conventional CHOP or CHOP-like regimens has significantly improved survival, approximately 30% of DLBCL patients fail chemotherapy [5]. The International Prognostic Index (IPI) based on age, performance status (PS), Ann Arbor stage, number of extranodal lesions and serum lactate dehydrogenase (LDH) level is a standard prognostic scoring system for predicting the clinical outcomes of patients with DLBCL. But in rituximab era, the ability of the IPI to predict prognosis has declined [6]. A variety of molecular biomarkers and gene signatures with prognostic significance have been identified in DLBCL patients [1–4]. However, molecular markers and gene signatures are expensive, technically challenging, and not routinely available in many undeveloped countries. Therefore, cheap and easily accessible prognostic markers which might help to increase prognostic accuracy are needed.

Malignant and inflammation are closely linked. Inflammatory processes have been identified to play an important role in the pathogenesis of lymphoma [7–9]. Pro-inflammatory cytokines (or chemokines) and inflammatory cells and in tumor microenvironment were proved to promote tumor growth, DNA damage, angiogenesis, immune suppression and to be associated with poor survival outcomes of patients [10–13]. Cytokine receptors or other factors in inflammatory pathways implicated in metastasis may be an appropriate target for malignant tumor therapy [14–19]. Circulating inflammatory parameter(including C-reactive protein (CRP), albumin levels, the lymphocyte-monocyte ratio (LMR), the neutrophil-lymphocyte ratio (NLR), the platelet-lymphocyte ratio (PLR) or fibrinogen levels) was associated with a poor prognosis in many types of malignant tumor patients including DLBCL [1, 20–27]. Several inflammation-based cumulative prognostic score systems were established for various solid tumors [28–31], but there is few inflammation-based cumulative prognostic score system for predicting survival of patients with diffuse large B cell lymphoma. Therefore, this retrospective study aimed to develop a novel inflammation-based cumulative prognostic score system for predicting survival of patients with diffuse large B cell lymphoma in rituximab era. The system may be useful for identify candidates for further inflammation related research and clinical trial of novel anti-inflammation drug.

It is well established that systemic inflammation status is associated with poor outcome in various solid tumors and lymphomas [5, 10, 33–36]. Prognostic parameters which reflect systemic inflammatory response include CRP, albumin levels, LMR, NLR, PLR and fibrinogen levels [1, 20–27], etc. Furthermore, recent studies had indicated that inflammation-based cumulative prognostic score systems, like the Glasgow prognostic score (GPS) or advance lung cancer inflammation index (ALI), are useful prognostic models for several solid tumors [28–31]. A retrospective study aslo showed that the modified Glasgow prognostic scores (mGPS) can be used as a predictor in diffuse large B cell lymphoma treated with R-CHOP regimen [37]: patients with lower mGPS had higher complete response rate and better OS, however, the difference of PFS was statistically insignificant. So far, there is few inflammation-based cumulative prognostic score systems for predicting survival of patients with diffuse large B cell lymphoma. To overcome the lacking data, we accessed prognostic values of inflammatory prognostic parameters commonly used in publications including CRP, albumin levels, LMR, NLR, PLR and fibrinogen levels. The results of univariate and multivariate analysis showed that albumin levels, CRP and LMR are three independent prognostic factors for OS. Based on these three independent inflammatory adverse factors (CRP > 8.6 mg/L, albumin levels < 41.5 g/L and LMR ≤ 2.7), we developed a novel systemic inflammatory cumulative prognostic score system, which we called the ICPS. An advanced multivariate analysis showed that the ICPS was a prognostic system independent of IPI risk factors including age, Ann Arbor stage, number of extranodal sites, ECOG PS and serum LDH level. Compared with any single inflammatory prognostic parameters, the ICPS model was shown to classify risk groups more accurately. The ICPS model can classify patients to four risk groups for OS and three risk groups for PFS. The survival difference between any neighboring two risk groups is significant in both PFS and OS analysis. As our best knowledge, it is the best inflammatory-based prognostic score system reported in DLBCL patients so far. Patients with higher ICPS had poorer clinical outcomes, which indicated that the degree of systemic inflammatory status had indeed been associated with clinical outcomes of patients with DLBCL in rituximab era. Inflammation plays an important role in malignant tumor development. Inflammatory cytokines (including tumor necrosis factor (TNF), Interleukins (IL) and chemokines, etc.) and immune cells (including tumor-associated macrophages and tumor infiltrating lymphocytes, etc.) in tumor microenvironment contribute to tumor growth, metastasis and immunosuppression [10–13]. Functional polymorphisms of inflammatory cytokine genes have been demonstrated to be associated with lymphoma risk and outcomes [38–40]. Chemokines or other inflammatory factors implicated in metastasis are considered to be potential targets for malignant tumor therapy [41–43]. Elevated levels of serum inflammatory markers reflect the inflammatory response of the body to malignant tumors. For example, C-reaction protein(CRP)is an acute phase protein produced by the liver after cytokines stimuli. Inflammation-based cumulative prognostic score systems were shown to be associated with levels of serum chemokines in patients with lymphoma [44]. The molecular mechanism of high degree of systemic inflammation status in malignant patients is unknown, whether these patients have abnormity of inflammatory pathways awaits future investigation. If be confirmed, these inflammation-based score systems may be useful in identifying patients for further inflammation-related mechanism research or novel anti-inflammation drug or gene target therapies.

At our lab, the normal range of CRP is 0-5 mg/L, the normal range of albumin level is >35 mg/L, the normal range of lymph:mono ratio is not defined. The cut-off value of albumin level in this study is 41.5 mg/L based on ROC analysis, which falls into the normal range. Univariate analysis showed that albumin levels were positively correlated with OS and PFS, regardless of whether the cut-off value was 41.5 mg/L or 35 mg/L. However, multivariate analysis with albumin level and IPI risk factors found that (data was not shown): when the cut-off value was set at 41.5 mg/L, low albumin level would be an independent risk factor for poor outcomes (P = 0.002 for OS, P = 0.004 for PFS). When the cut-off value was set at 35 mg/L, the prognostic value of low albumin level disappeared (P = 0.442 for OS, P = 0.338 for PFS). Although albumin level 35~41.5 mg/L was in the normal range, survival of those patients were no difference with whose albumin level < 35 mg/L (3-year survival 74.7% VS.67.5%, P = 0.220 for OS; 69.4% VS.59.7%, P = 0.091 for PFS), but significantly worse than whose ≥ 41.5 mg/L(3-year survival 74.7% VS.92.1%,P < 0.001 for OS; 69.4% VS.83.8%,P < 0.001 for PFS). So we defined 41.5 mg/L as cut-off value other than 35 mg/L. Most other studies of DLBCL patients used 35 mg/L as a cut-off value, which may lead to negative results. We speculate that slight decrease of serum albumin level may be caused by increase of inflammatory cytokines secretion and vascular permeability which is induced by systemic inflammatory states in lymphoma patients. While severe hypoalbuminemia may be caused mostly by malnutrition, tumor consumption, pleural or peritoneal effusion and edema, so it is independently associated with IPI risk factors. This hypothesis requires more evidence to support.

There are several limitations in this study. First, some cases chose treatment regimens without Rituximab or did not complete their course of R-CHOP treatment because the cost of Rituximab was beyond which they can afford. In order to ensure the homogeneity of treatment, these patients were excluded, which might result in a selection bias. Second, the ICPS model is based on a single-center retrospective study in South China, which has not been validated in other populations. Multicenter prospective studies are needed to determine whether it is suitable for large sample groups.

In summary, we have developed a novel systemic inflammatory cumulative prognostic score system independent of the IPI in DLBCL patients treated with R-CHOP therapy. The score system based on serum CRP, albumin levels and peripheral LMR, which we called the ICPS. The ICPS model can classify patients into four risk groups with significantly different survival outcomes. The research indicated that the degree of systemic inflammatory status was associated with clinical outcomes of patients with DLBCL receiving R-CHOP therapy. Additional prospective multicenter studies are needed to confirm the clinical potential of the ICPS as prognostic system in patients with DLBCL. The molecular abnormity of high degree of systemic inflammation status in these DLBCL patients awaits future investigation. The ICPS model may be useful in identifying patients for further inflammation-related mechanism research or novel anti-inflammation gene or drug target therapies.